"Face it: You are simply trying to defend your CDN business as it is being bypassed by direct connections between content providers and last mile ISPs."

Its clearly a telecom employee that wrote that or someone who is ignorant. If they weren't, they'd realize https://www.netflix.com/openconnect was Netflix offering to do that for free.

Netflix is paying to get the data from their datacenters to the LEC regardless.

Doesn't this cease to scale at some point?

At the moment netflix is 50% of internet bandwidth.

If that could be delivered from a CDN within the ISP, that takes a majority of traffic off the congested pipes and leaves enough room for all the current demand with quite a bit of overhead.

> If that could be delivered from a CDN within the ISP, that takes a majority of traffic off the congested pipes and leaves enough room for all the current demand with quite a bit of overhead.

For how long?

Every year the hardware vendors come out with faster equipment which makes bandwidth cheaper which causes people to use more. If they won't upgrade the interconnects then all removing Netflix does is buy a little time until the interconnect gets saturated again. When that happens you can't remove the Netflix traffic again, so someone else will be forced to pay, who will be smaller than Netflix and therefore their paying the danegeld will buy even less time for everyone else.

Meanwhile, providers pocket the extra money from 20 years of Moores law while nobody on the last mile benefits. It's pretty clear that in the last decades, running an ISP can only have gotten ridiculously easier. And here they are not wanting to provide interconnect bandwidth.

And the 50mbit domestic connections wasn't at all common 20 years ago (heck, we did have 33.6kbps dial-up and it cost a fortune). In fact, if you say that there haven't been large, fast changes in bandwidth terms then I'd invite you to revisit the following facts:

- Smartphones in every pocket, with people consuming 1-2 GB a month. This was unthinkable 7 years ago — and I vividly recall spending a ton of money because I used 1 GB of mobile internet. - 4G in every corner — just a few years ago the top was the 384kbps UMTS - FTTH and DOCSIS 3.0 ubiquitous, mainly to lowered costs for fiber equipment, GPON lowers the cost and tougher fiber cables that can easily be installed in houses - Ten-fold growth in internet provides bandwidth… or more. For example, <10 years ago, OVH's network (one of the largest dedicated server providers in the world) was one of the largest at 300Gbps (I remember the party when they reached out that capacity!). Now they are at +3Tbps and growing. - 1080p and 4K increasingly common — and then a decade ago we were still buying CRT televisions, a far cry from the FHD and higher bitrate contents

I could keep going with the list… But the fact is: everywhere you look bandwidth is getting cheaper AND faster, same goes for consumer devices that profit said bandwidth (TVs, smartphones…) and that comes in hand with more, richer content.

Yes, peasant in Romania is able to get 1Gbit FTTH for 20 Euro in rural nowhere while at the same time hipster in downtown NY will be happy to get 30Mbit bundled with shitty cable at $150.

It is more like 32%. The 50% number was all video.

I don't see how the former is better than the latter from a net neutrality point of view. In both cases, Netflix's humongous traffic is being treated as a special case, and in both cases Netflix is paying to deal with it.

Further, netflix watchers are almost entirely composed of former/would-be TV watchers who ultimately end up spending less time in front of the tube because they avoid commercials entirely (adding up to a ton of time in aggregate) as well as at their leisure instead of when the TV guide demands they stop what they're doing to catch their show.

http://www.nydailynews.com/entertainment/tv-movies/americans...

People spend more time watching TV as they do working on average. [34 + 3 = 37]

http://www.bls.gov/news.release/empsit.t18.htm

34.5 hour work week average

Netflix isn't shoving their packets down anyone's throat here, the ISP's whining notwithstanding.

Pizza delivers to me. Amazon delivers to me. NewEgg delivers to me.

Please provide a counter example that implies what you suggest? All of these imply the first noun delivers something requested by the second noun. I'm uncertain why that is unclear?

ISPs already allow businesses to co-locate servers, netflix could have done the same. But it'd have to at least pay colocation fees + bandwidth fees.

That said, I wouldn't be surprised if Netflix will/does pay some amount to colocate in some instances. The economics in some instances may cause the amount paid to be less than the cost of the CDN traffic.

You are ignoring the fact it is offering this to save the ISPs money and improve the quality of the service.

Its cheaper for the ISPs to put an openconnect box inside their network in the LA area than it is to pay for more networking equipment.

ISPs are expected to provide sufficient capacity at peering points with other networks to provide reasonable quality delivery to their customers.

How they handle that without demanding special privileges (as ISPs are in Netflix's case by holding traffic hostage until they are paid more money) is their buisness.

DSL (when not of the ATT U-verse variety) barely gets over 10-Mbit in practice, with speeds closer to 3-Mbit being fairly common.

Satellite can't really be considered high-speed due to latency.

LTE coverage is still poor, and I've yet to see hardware suitable for a home connection for it. (Tethering via a cellphone doesn't count.)

So, no. There's really generally one, maybe two choices for home high-speed internet so it's not like customers can vote with their dollars and move to a Netflix-friendly ISP.

Well, they don't get free access to the lines, they still pay rates, typically, from what I understand around 15-35% lower than the usual consumer prices per customer to the "main" provider on that line. If that's really not enough for them to continue maintaining and improving their infrastructure perhaps it would be an issue - but it certainly appears to have been sufficient here at least.

Satellite has caps and active throttling which prevent video streaming, so no Netflix for you.

On DSL, if your sync rate is 3Mbps, you won't be watching anything in HD and unless you dedicate the line for only Netflix and ban everybody else in the house from using the internet, you'll get poor quality SD only.

So, no, DSL and satellite are not plenty fast enough for Netflix.

This is an issue of customers not being able to use what they purchased, eg The Internet.

The water was arguably muddy when they were advertising "Speeds up to X", but now when their own infrastructure is incapable of providing what they're selling and they are fully aware of this, they no longer have a leg (or argument) to stand on.

Either way you cut it, the ISP is in the wrong. The peering point is 100% within their control, and Level3 has gone above and beyond to try to ease congestion, the ISP has outright refused.

How on earth can you side with the ISP on this - do you work for one?

Comcast? Comcast and Verizon were the big two Netflix signed deals with.

http://mashable.com/2014/07/29/netflix-att-peering-deal/

Would be a booming startup for sure. Very surprised google has not taken this one on. I hate being forced to have only once choice for internet (cox). You know that there is not much of incentive for them to do a good job when they have no competition!!!!

This problem is a fundamental issue of RF Spectrum, and these problems tend to be addressed by blue sky research that occurs at a university.

Google is trying to address this in the most cost effective manner possible... google fibre. Wireless is unlikely to be any cheaper then simply rolling fibre / copper to every home.

4 cities available in california for google fibre!! That is nuts. I hope they move a lot quicker or someone else should.

PS: This is why you can have temperary cell towers for events. http://en.m.wikipedia.org/wiki/Cell_on_wheels

http://www.satbroadcasts.com/DVB-T_Bitrate_Calculator.html

And:

http://en.wikipedia.org/wiki/DVB-T2#Technical_details

?

What makes you think that?

Anecdote: I live in a new building in Baltimore, and have both Comcast cable and Verizon FiOS. The FiOS is an MDU, so instead of fiber to each apartment we get fiber to an ONT on each floor, and VDSL to each unit. Still, it's rock solid 45/15 all times of day for $50/month.

The other day, I lost internet. Called Verizon, they said it must be a problem with my unit because nobody else in the building had complained. Verizon tech comes out, and realizes that the reason nobody has complained is that I'm the only one on the ONT. Indeed, the ONT wasn't even properly configured with the network, and nobody had noticed because I was the first person on the floor to subscribe to the service since the building was built in 2011. We're talking about a building with dozens of apartments on each floor, by the way.

The moral of the story is that people care about their internet connection a lot less than folks on HN would like to believe. Verizon is still pushing hard to recoup its investment in FiOS, because customers aren't willing to pay what it costs to build all that infrastructure. Why would a startup want to enter that market?

  In 1994 Verizon (then Bell Atlantic) struck a landmark deal with the state of Pennsylvania. The deal provided Verizon with hefty financial incentives if they met certain broadband rollout criteria. It's estimated that those financial incentives over the years clock in somewhere around $2.1 billion dollars.

  As part of that agreement, Bell Atlantic agreed to have 20% of the state broadband wired by 1998, and 50% by 2004. By 2015, broadband would be run throughout the state to the majority of Verizon's customers. It's important to note that this wasn't DSL they were talking about...but 45MB/s symmetrical fiber service right to the door of homes and businesses, ambitious and impractical for certain, but nonetheless included in the language of the agreement. [1]

Where does the $2 billion come from? It starts on page 4 of the document.

> Customers paid for a network they will never receive. We estimate that the Company received $2.1 billion from this deregulation, including an additional $1.5 billion in extra tax deductions the Company received from excessive write-offs of the still existing networks.

He gets into the real numbers starting on page 23.

The $2.1 billion is Kushnick's usual fabrication: pick a "rate of return" typical for regulated monopolies, and call all profits above that number post-deregulation as "money given to the telephone company."

Page 28+ gets into the tax angle. The basic premise of his argument is that Bell Atlantic took an unjustified accelerated depreciation on its copper phone network. He paints this as a scam: they took the write-off on the basis that they were going to replace it all with fiber, then failed to do so.

However, that wasn't actually the accounting justification for taking the write-offs.

> In such markets, the Company does not believe it can be assured that prices can be maintained at levels that will recover the net carrying amount of existing telephone plant and equipment, which has been depreciated over relatively long regulator-prescribed lives.

In other words, the accelerated depreciation was to compensate for the fact that once regulated prices were gone, the telephone plant would become less valuable. It was not predicated on replacing that plant with fiber.

At best we're talking about Bell Atlantic taking depreciation deductions faster than they should have. It's certainly not equivalent to Pennsylvania writing it a $2 billion check for fiber that was never delivered, or even giving them $2 billion in tax deductions for fiber that was never delivered. Maybe this accelerated depreciation was not proper. I'm not a tax expert. But neither is Bruce Kushnik.

The scam is simple -- by using title II, they get to charge local phone customers rate increases to pay for the fiber optics. Verizon also dumps the construction budgets into the utility networks for its wires to the cell towers and even the 'special access wires' -- which are also classified as Title II. – Title II, then is a cash machine.

And Verizon never told the FCC, the courts, or the public about this.

But the revenues don’t go back to the utility—they appear to go into a “black hole” accounting…

In tracking Verizon New York, we found that there were actually statements made about raising rates for the ‘massive deployment of fiber optics’ in 2009, which was the third increase since 2006.

The increases, including the additional taxes, fees and surcharges (many of which are also either revenues to Verizon or ‘pass-through’ taxes that are on Verizon but charge customers), and the increases to all calling features, etc – came to about $4.5 billion extra since 2006.

However, the data – Verizon stopped publishing its SEC-based state reports in 2010, the FCC stopped publishing the data in 2007, but NY State required an annual report – but it only shows the revenues and expenses for the utility.

The ‘black hole’ funds were uncovered when comparing the state-based SEC reports with the State-filed annual reports; in just New York, in just 2009, there was an additional $2.7 billion in revenues – but no extra construction costs to this ‘black hole revenue.

Verizon then, shows losses in the utility, claiming the networks are uneconomical to upgrade. But the losses are created based on this flow money—the ‘affiliate’ companies, like Verizon Wireless or Verizon Online or Verizon Business, pay less than market prices and have the construction budgets dumped on the regulated side—lowering revenues and adding expenses and creating manipulated losses.

Verizon NY alone showed about $11 billion in losses, about 2 billion a year from 2008…about $5 billion in tax savings… Ie, Verizon New York paid no income taxes since 2008 or earlier.

This is happening in every Verizon state, and we assume AT&T as well but they aren’t required to supply basic data anymore. AT&T stopped publishing its SEC reports a decade ago or more, and most, if not all states, don’t require the level of financials needed to examine this flow of money.

This new financial shell game started to pick up speed after the networks were closed to competition around 2005.

So, on top of the original ‘commitments’ and the extra money that’s been collected since the 1990’s, as no state ever went back and got the money or refunds to even stopped the excess profits, this new financial-game puts the original overcharging customers as a very low number as it doesn’t account for the steriod-based, Title II, cash machine.

First, I have a new book which details the state materials. “The Book of Broken Promises” and goes through 2014, and covers most of the state commitments to deploy fiber optic broadband, monies collected and the failure to deliver and goes through 2014.

  http://newnetworks.com/bookofbrokenpromises.htm  

>The $2.1 billion is Kushnick's usual fabrication: pick a "rate of return" typical for <regulated monopolies, and call all profits above that number post-deregulation as <"money given to the telephone company."

I didn’t pick the rate of return. Duh- It was based on the current returns that were allowable under state law, and I used the PA numbers—as told by the state-based Bell of PA (now Verizon PA) SEC filings.

The changes in state law happened in every state (though with varying commitments) – such as Verizon states -- PA, NJ, MA, and other states, IL, OH, etc. -- and they were ALL based on direct commitments to rewire parts, if not the entire states as well as schools, depending on the state.

In PA, the original commitment were 100% completed by 2015, (though the commitments were changed over time.) in NJ it is 100% completed by 2010—with fiber 45 Mbps in both directions.

In PA we filed a complaint which said – state laws were changed specifically based on the commitments to upgrade the state utility plant, which was copper, with fiber. It removed the old ‘rate of return’ for new ‘alternative regulation’, that speeded up the write offs of the networks, and no longer examined the profits of a state utility – which had a monopoly on that wire—with the goal to use this extra money for the deployment plan called Opportunity PA.

The profits overall, went from 12%-14%, which was standard in almost every Bell state to about 30% after the new law took effect. In PA it went from 13% then jumped to 30% by 1998. -- But they didn’t deploy the fiber – nor replace the wire. And these numbers were in the SEC-annual reports.

You didn’t bother noting the chart on page 23 outlining the extra profits—directly tied to deploy fiber optics.

And PA and NJ had a timeline of deployment to upgrade to fiber and 45 Mbps—and it didn’t happen. An analyst we quote said:

“As we approach the end of 1998 a point by which BA-PA is supposed to have broadband available throughout 20% of its rural, urban and suburban areas there is no sign of any broadband service being offered to Pennsylvania's residential customers."

And the PA state commission wrote:

“When the Commission accepted Bell's proposal, that proposal became binding on the Company. Any modifications or deviations from a 45 Mbps two way interactive network must be approved by this agency, since such would constitute a modification to the June 28, 1994 Opinion and Order which ruled on the Company's original Petition and Plan.”

You write: >In other words, the accelerated depreciation was to compensate for the fact that once >regulated prices were gone, the telephone plant would become less valuable. It was not >predicated on replacing that plant with fiber.

Wrong… it was to replace the copper wires in the state utility. You obviously didn’t read the documents. This is the language in the New Jersey Bell (Verizon NJ) annual report, 1994, for a $1.013 billion deduction – for the “company’s technology deployment plan” — It was a one-time deduction which every phone company took, called FAS 71, and it says:

"The Company's determination that it was no longer eligible for continued application of the accounting required by Statement No. 71. It was based on the belief that the convergence of competition, technological change (including the Company's technology deployment plans).

And the result is a tax savings of about ½ billion dollars in NJ —from this one change.

And this is on top of ‘accelerated’ depreciation, which was also set by the state commissions to help speed up the deployment to fiber optics—which didn’t show from 1993-2005. – i.e., the company gets more tax deductions per year and didn’t replace the copper.

We documented this in our first book, 1999, Foreword by Dr. Bob Metcalfe, (co-inventor of Ethernet) who went through all the stats. And it was ALL for the commitment to replace the copper wires with fiber.

By around 2009- in NJ we calculated that Verizon NJ collected about 15 billion.

And you’re apologist for the companies deceiving the public and the state commissions and getting billions extra in tax perks and excess profits – which are monies that directly impact customers’ bottom line.

Law changed – extra money garnered – nothing built, (1993-2005) but to you that’s just fine.

And this PA stuff is from 2002-2003, before we uncovered the entire story about Verizon and AT&T financial manipulations—and it was low number.

Anyone interested in knowing the real history about broadband and fiber should do the fact checking themselves – the new book has detailed footnotes and links.

There is competition to Cable and DSL for Internet access, but most operators avoid the low-margin retail space they operate in.

[1] - https://youtube.com/watch?v=wGAnDQEQJ_s

https://www.youtube.com/watch?v=4eMBBVG-MNY

http://www.tarborotimes.com/2014/04/26/weird-news-google-buy...

Current experiments are putting the performance at the 3G level, but it seems likely only a matter of time before these have LTE or better.

As far as I've been able to find out, their backhaul is some sort of wireless link to their NOC.